Temporal Variation of Atmospheric Static Electric Field and Air Ions and their Relationships to Pollution in Shanghai

Atmospheric electric field (EF) and air ions were measured in Shanghai from December 2014 to December 2015 to examine the influence of particulate pollutants. Fair-weather EF exhibits a diurnal variation of multi-modal oscillations in spring, summer, and autumn. Linear correlation analyses show that the local meteorological conditions of relative humidity, temperature, pressure and wind affect atmospheric electric field, with wind direction exhibiting the highest correlation coefficient. Atmospheric EF is significantly higher in the west compared with that in the east, as the air mass from inland areas carries more polluted aerosols. Air ion concentrations are generally higher in the daytime than at night and correlate with meteorological factors. Atmospheric EF undergoes a substantial fluctuation in polluted periods, but remains flat under clean conditions. Overall, in areas with pollution, the atmospheric EF gradually increases with increased pollution or increased particle loading (e.g., PM2.5), a useful indicator of air pollution. The concentration of PM2.5 is positively correlated with the atmospheric electric field under polluted conditions, because as the concentration of aerosol particles declines, the concentrations of small ions and the atmospheric conductivity decrease accordingly, thus causing an increase in the electric field. Overall, aerosol particles, air ions, and their interactions in the presence of various meteorological parameters can have local effects on the atmospheric electric field.


INTRODUCTION
An electric potential difference of about 250 kV between the ground and the ionosphere produces a vertical downward atmospheric electric field (EF) at the surface of the earth.The ground-level vertical component of the EF may be divided into global and local components which mainly depend on the location of the measurement station and the local time.The global components include ionospheric potential, solar radiation, and natural radioactivity, and local components include atmospheric conductivity, space charge density, aerosols, and meteorological conditions.
Numerous measurements of the intensity of surface EF have been conducted at marine, continental, polar and mountainous areas to explore global or local effects of atmospheric EF (Deshpande and Kamra, 2001;Harrison, 2013;Siingh et al., 2013;Kamogawa et al., 2015).
The atmospheric electric field is an important parameter used to characterize the global electrical circuit (Harrison, 2006;Aplin, 2012;Rycroft et al., 2012).Generally, atmospheric EF shows a single intra-day oscillation over marine and polar areas where local effects are absent or low, in an effect referred to as the Carnegie curve (maximum around 19:00 UTC and minimum around 03:00 UTC) (Harrison, 2004;Harrison, 2013).Whipple (1929) presented a correlation between the Carnegie curve and the diurnal variation of the global thunderstorm area.The timing of the thunderstorm activity between the peaks in the values of electric field have a strong positive correlation, and the minimum in the Carnegie curve correlated with low lightning activity over the Pacific at 4 UT (Williams and Sátori, 2004).Therefore the Carnegie curve mainly reflects the global effect, and it can be in marine and polar areas where local effects are absent or low.However, there are small differences in this pattern at a few sites, such as Maitri of Antarctica (Deshpande and Kamra, 2001).Aerosols can cause large variability of EF at continental sites by attaching or neutralizing small ions to reduce the air electric conductivity (Harrison, 2006;Silva et al., 2014;Conceição et al., 2016).Many studies have demonstrated the relationship between aerosols and atmospheric electricity.Adlerman and Williams (1996) reported a nonlinear relationship between aerosol and electrical conductivity over land.A positive correlation was found between the electric field and aerosol concentration (Jayaratne and Verma, 2004).Sheftel et al. (1994) showed that atmospheric EF and air conductivity are good indicators of anthropogenic atmospheric pollution.Kubicki et al. (2007) showed that the maximum annual variation of EF in winter resulted from high aerosol and dust concentrations at a land station.
Air pollutants such as nitrogen oxides (NO x ), sulfur dioxide (SO 2 ), ozone (O 3 ), carbon monoxide (CO) and particulate matter (PM) significantly, affect air quality (Chelani, 2013;Wang et al., 2016).Recent studies revealed a high correlation between EF and air pollutants (Martín et al., 2003;Silva et al., 2016), and this effect was much stronger in winter.Air pollutants increase the ion recombination rate, causing a decrease of net ion concentration that reduces the air conductivity and increases the EF.Srivastava et al. (1972) observed that changes of EF and conductivity in the planetary boundary layer (PBL) are closely associated with dust, haze, and the local meteorological conditions.Piper and Bennett (2012) found that shallow convection and radiation fog can obviously influence the near-surface EF.
Atmospheric ions are mainly generated by the ionization of air molecules by ground and galactic cosmic radiation and radioactive materials such as Radon (Rn) and their short-lived progeny (Cheung et al., 2015;Li et al., 2015).Small ions are usually called molecular cluster ions, less than 1.6nm in size, and their recombination or attachment to aerosol particles allow to growth into intermediate (1.6-7.4 nm) and large (7.4-79 nm) ions (Pawar et al., 2010;Li et al., 2015).Additionally, the ion mobility decreases with increasing size (Tammet, 1995).Many studies have studied atmospheric ions and their contribution to particle formation (Harrison and Carslaw, 2003;Wilding and Harrison, 2005;Cheung et al., 2015;Jayaratne et al., 2016).Atmospheric ions also play a decisive role in atmospheric electrical variation.Pawar et al. (2005) measured air ions and found that large ions strengthen atmospheric electric conductivity.Harrison and Carslaw (2003) investigated the influence of atmospheric charged particles on aerosol and cloud microphysical processes, and examined the connection between atmospheric ionization rate and global cloudiness and weather systems.
Because they carry many electric charges, water droplets and rain-clouds can change EF drastically within a short time, especially during thunderstorms (Bennett and Harrison, 2007a).We screened out extreme weathers and only focused on fair weather to study near-surface atmospheric EF.We measured atmospheric EF and air ions in Shanghai from 2014 to 2015 to characterize their temporal variations and relevant influence factors.This information allows insights into the correlation of these parameters with pollution in urban environments with significant anthropogenic emissions of air ions.

Site Description
The observation site is located on the roof of a building about 20 m height above the ground on the campus of Fudan University (31.29°N, 121.51°E) in Shanghai of China, 40 km east of the East China Sea, surrounded by urban commercial and residential areas (Cheng et al., 2015).The industrial emissions primarily come from a nearby steel plant and a power plant, and the nearest industrial sources are approximately 10 km away, primarily located in the northwest and the southeast (Li et al., 2011).

Theory
In fair weather without intense atmospheric convection, according to Ohm's law, air electric conductivity (σ) is positively proportional to the current density (J), but inversely related to the atmospheric electric field (E).
Air electrical conductivity mainly depends on air ion concentration and mobility, especially for small ions, and the equation is as follows: (2) Specifically, n + and n -refer to the concentrations of positive and negative ions, µ + and µ -refer to positive and negative ion mobility, and e is the elementary charge.If bipolar ion concentrations are equal, the above equation can be simplified.Air ions are produced by galactic cosmic radiation and radioactive materials and are removed by recombination or attachment to aerosol particles.The stable state can be described as follows: where q is the ion production rate, α is the ion-ion recombination coefficient, β is the ion-aerosol attachment coefficient, and Z is the aerosol number concentration.In case of severe pollution, aerosol particles at high concentration can result in increased ion-aerosol attachment to favor ion removal, i.e., nβZ ≫ αn 2 , and then Eq. ( 3) can be simplified into Eq.( 4).n = q/βZ (4) Substituting Eq. ( 4) in the equation Eq. ( 2), and applying those results to Eq. ( 1), we obtain Eq. ( 5).

Instrumentation and Observation
An electric-field mill (EFM-100, Boltek Corporation) utilizes the conductor in an electric field induced charge to measure the EF intensity at the surface of the earth.The electric-field mill was calibrated by the manufacturer with a parallel plate method before our use of the instrument for observation.The output signals were digitized and recorded every 0.5 seconds over the range of ± 20.0 × 10 3 V m -1 .It is very difficult to find an open place for observation, so the electric-field mill was installed on the roof of a barrierfree building.The EF on the building roof is stronger than on the ground due to the distortion effect caused by the surrounding cusp or steel structure building (Zhou et al., 2010;Tan et al., 2014).
The EF is often about several hundred V m -1 in fair weather, and becomes larger or negative in disturbed weathers such as thunderstorms, precipitation, and clouds (Bennett and Harrison, 2007a).In order to eliminate the effect of bad weathers, we collected data only for days with good weather (wind speed < 6 m s -1 , cloud coverage < 3/10 and no significant rainfall).This provided data for 54 days.No data was collected in February or November 2015 due to instrument maintenance.
An air ion mobility analyzer (DLY-71, Kilter Electronic Institute Co. Ltd., China) was used to measure air ion concentrations at the surface of the earth, with a magnitude resolution of 10 ions cm -3 and time resolution of 30 min.The instrument is able to measure small (< 1.6 nm), intermediate (1.6-7.4 nm) and large (7.4-79 nm) ions of both polarity (negative or positive), with mobilities of 3.2-0.4,0.4-0.04 and 0.04-0.001cm 2 V -1 s -1 , respectively.
Additionally, a set of micro pulse lidar (MPL, Sigma Space Corporation, USA) was applied to monitor the vertical profile of aerosol backscattering at 532 nm.The height of the planetary boundary layer (PBL) was obtained from MPL measurements using the classical Fernald algorithm and a lidar ratio of 30sr.The PM 2.5 concentration was measured using an on-line particle mass analyzer (Thermo SHARP-5030).A visibility monitor (VPF-730, Biral) was utilized to measure atmospheric visibility.Meteorological variables were measured at an automatic weather station (HydroMet™, Vaisala).

Fair-Weather Atmospheric Electric Field
Fig. 1 shows the frequency distribution of hourly-averaged EF intensity in fair weather.The hourly EFs varied from 0 to 2 kV m -1 , with an average of 692 V m -1 .As show in the histograms, the hourly EFs were mainly concentrated in 500-800 V m -1 , similar to a Gaussian distribution with a fitting correlation coefficient of 0.98.The mean EFs were about 630, 660, 620 and 820 V m -1 in spring, summer, autumn and winter, respectively (Table 1).Higher EFs in winter and lower EFs in the summer were observed and are presented in Table 1 and Fig. 2.These results agree with those from other continental sites in the northern hemisphere (Retalis and Retalis, 1997;Israelsson and Tammet, 2001;Kastelis and Kourtidis, 2016).A possible explanation for this seasonal variability is likely increased aerosol concentration near the surface during winter, because the lack of strong surface heating and deep convective mixing means that pollution generated at the surface is confined to the nearsurface where the EF is measured and not distributed upwards by summertime convective mixing (Bennett and Harrison, 2007b).
The hourly EFs show an obvious pattern of diurnal variation with multiple peaks (Fig. 2).Overall, the EFs were higher in the daytime than at night, especially in winter (> 1000 V m -1 ).Generally, the EF peaks in winter appeared at 08:00-11:00 LT and 13:00-16:00 LT, and between these two peaks, the latter one was much higher.The EF peaks in the other three seasons occurred at 06:00-09:00 LT, 13:00-15:00 LT and 17:00-19:00 LT.The variation of the peaks in the daytime is possibly related to the effects of sunrise and sunset and human activities (Retalis and Retalis, 1997), particularly in spring, summer and autumn.At night, similar to the data from other continental stations, the higher EFs are likely due to the decrease of exchange layer heights (Latha, 2003).The bi-and multi-modal patterns were also observed at other continental sites (Harrison and Aplin, 2002;Kastelis and Kourtidis, 2016), and are largely affected by local factors of aerosols, meteorological conditions, space charge, and natural radioactivity.The atmospheric EFs also display a mono-modal pattern of diurnal variation at pristine places, such as in polar, oceanic, mountainous, and remote regions (Deshpande and Kamra, 2001;Israelsson and Tammet, 2001;Harrison, 2003;Panneerselvam et al., 2007;Harrison, 2013;Siingh et al., 2013;Xu et al., 2013;Kamogawa et al., 2015;Kastelis and Kourtidis, 2016).
The sunrise effect has been observed at tropical continental, rural and island areas (Latha, 2003).Law (1962) and Kamra (1982) reported that the EF increase is the response to the sunrise, positive space charge density increase and  nocturnal inversion breakdown, but the conductivity rarely varies.Marshall et al. (1999) found that the sunrise effect is related with electrode layers and space charge transport.Additionally, morning solar heating leads to the dilution of the electrode layer built up during the night.More positive charges aggregate as a result of increasing vertical turbulent transportation and eventually give rise to EF increase (Marshall et al., 1999).In our study, aerosol particles were positively correlated to EF, indicating that in addition to the sunrise effect, the influence of human activities on EF can be significant in the morning, especially in winter.In addition, the meteorological conditions are local influence factors on EF at the surface of the earth on smaller scales, particularly over polluted areas.The correlation and variance of EF and meteorological parameters are summarized in Table 2.
Overall, the meteorological parameters showed a weak correlation with EF (Sig < 0.05), with wind direction the most correlated factor (0.54), and wind speed the least correlated factor (0.3).Generally, northwest, east, and west winds prevail in fair weathers, and the EFs mainly distribute in the west and east directions (Fig. 3).The EFs is lower in the east wind than the west wind, about 145 V m -1 .A possible reason for this is the east wind from oceanic areas carries clean air, but the west wind from inland areas carries more pollutants (Wang, 2013).

Air Ion Concentrations
Fig. 4 presents frequency histograms of the hourly positive and negative air ion concentrations, including large (L + , L -), intermediate (M + , M -), and small (S + , S -) ions.All ions present a similar Gaussian distribution, with fitting correlation coefficients of about 0.82-0.96.The L + and L - ions are mostly in the range of 1500-2000 cm -3 , S + and S - ions are in the range of 90-150 cm -3 , and M + ions are in the range of 20-60 cm -3 and M -ions are in the range of 40-100 cm -3 .Table 1 summarizes the annual and seasonal averages of air ion concentrations in fair weather.Overall, the amount of large ions per volume is more than 20 times that of intermediate and small ions, and intermediate ions are the most rares.The mean concentrations of L -and S - are relatively higher in winter, autumn, and spring, and L + and S + are higher in winter, autumn and summer.The mean M -concentrations are relatively higher in winter, and M + is higher in the winter and spring.
The correlation coefficients of the positive and negative ions were determined as 0.4, 0.71, and 0.35 for the large,   intermediate and small ions, respectively (Fig. 5), indicating that are not derived from exactly the same sources, such as vehicle emissions in an urban environment (Cheung et al., 2015).Each ion is assumed to be equally charged, and atmospheric large, intermediate, and small ions do not neutralize with each other.The electric polarity of large and small ions is usually negative in winter and spring (L -/L + or S -/S + = 1.0-1.9),and positive in summer and autumn (L - /L + or S -/S + = 0.9-1.0).The intermediate ions show negative polarity in winter, summer, and autumn (M -/M + = 1.1-1.3)but positive polarity in summer (M -/M + = 0.86).Overall, in terms of measured ions, the atmosphere shows weak negative polarity in winter and spring, and weak positive polarity in summer and autumn.Fig. 6 depicts the diurnal variations of hourly air ion concentrations.Overall, air ion concentrations are relatively higher in the daytime compared to those at night, possibly due to the ionization effect by solar radiation.The L + and S + begin to rise at 6:00 LT, and subsequently reach the first peak at around 9:00 LT, the second peak at around 13:00 LT, and the third peak around 20:00 LT before reaching a minimum at midnight.Similarly, L -and S -peaks occur around 10:00, 16:00, and 19:00 LT, but the L -and S -peaks are lower and last longer than those for L + and S + .For the intermediate ions, the M -concentration is almost larger than the M + concentration, and their peaks appear during 5:00-8:00, 10:00-15:00 and 19:00-20:00 LT.In terms of polarity, the large and small ions show positive polarity at around 0:00-4:30, 8:00-9:00 and 12:00-14:30 LT, but negative polarity dominates during the other times.However, the intermediate ions show mostly negative polarity throughout day.
The meteorological conditions influence the positive and negative ions (Fig. 7).As temperature, visibility, and wind speed increase, the large, intermediate and small ions become denser per volume, but the opposite trend is observed when the relative moisture rises.Moreover, the intermediate ions exhibit a greater sensitivity (i.e., slope) to varying meteorological parameters than the small and large ions.Temperature increases will strengthen the ionization process to increase the concentration of ions.However, a high relative humidity, a high concentration of amount of water molecules per unit volume, may cause a high collision probability and an increased combination rate between water molecules (Retalis et al., 2009;Hirsikko et al., 2011;Li, 2015).Therefore, the concentration of air ions is higher under conditions of low relative humidity or high temperature.
Fig. 8 presents the wind rose plots of positive and negative air ions.In brief, since the wind direction and wind speed can affect the transportation of pollutants, especially particles,    the air ion density changes accordingly.Positive ions prevail in the southwest (WWS), northwest (WN), north (N), and southeast (EES), and negative ions are prominent in the southwest (WWS), northwest (NWN), northeast (EN), southeast (EES) and south (S).In contrast, positive ions have a relatively weaker dependency on wind directions than negative ions, consistent with previous results (Cheung et al., 2015).
In these result, there was a greater influence of the wind on air ions compared to the effect on atmospheric EF, but this finding may reflect the lower amount of available air ion data as a result of low time resolution (30 min).

Atmospheric Electric Field and Air Ions in Pollution
Aerosol particles can attach to or neutralize small ions to decrease ions density and conductivity, thus increasing the atmospheric electric field (Law, 1962;Kamra, 1982).Air ions involved in particle formation and growth by recombination or attachment in an urban environment (Cheung et al., 2015).In fact, the atmospheric electric field is linked closely with pollutants, especially particles and their precursors, and can be viewed as an indicator of atmospheric pollution (Sheftel et al., 1994;Guo et al., 1996).Here, we analyzed the variation of the electric field, air ions, and pollutants in pollution.Fig. 9 shows a time series of atmospheric EF, air ion concentrations, and meteorological parameters from 11 to 13 October 2015.During this period, the PBL height was close to 500 m, the relative humidity was close to 60%, atmospheric visibility gradually decreased, and PM 2.5 increased from 23 to 139 µg m -3 , indicating that the air pollution worsened over this time period.In addition, despite fluctuation, the EF showed a slow growth trend (Fig. 9(c)).There were no obvious change in the distribution of the large, intermediate, and small air ions (Figs.9(d)-9(f)).
On the 11th of October, the PM 2.5 was measured at 54 µg m -3 , the PBL height was about 567 m, and the EF waves were about 542 V m -1 .In the morning, due to the sunrise effect and strengthened air vertical turbulent and transportation, the PBL height increased from 300 m to 800 m.More ions, especially positive ions were generated with increased amounts of aerosols.At our measurement site, human activities such as motor vehicle emissions and electrical appliances like transformers and air conditioners are the major sources of atmospheric ions and charged particles (Jayaratne et al., 2016).Under natural, stable conditions, atmospheric ions are present at concentrations of about 300-400 cm -3 but the concentration may increase to a few thousand cm -3 in the presence of anthropogenic ion sources such as motor vehicle exhaust (Maricq, 2006;Ling et al., 2010) and overhead power lines (Fews et al., 1999;Jayaratne et al., 2008).In this study, a higher concentration of positive and negative ions was found as shown in Fig. 9.The higher amount of positive ions relative to negative may be explained in terms of the higher mobility of negative ions.Negative ions have a higher efficiency of ion neutralization and, therefore, their lifetime and concentration are generally lower than those of positive ions (Kolarž et al., 2009).The lifetimes of small ions are significantly reduced in more polluted environments because ions with higher mobility (negative) will be neutralized faster due to their ability to attach to particles (Jayaratne et al., 2016).Therefore, the concentration of positive ions is higher than negative ions as a result of growing aerosols from human activities.The first peak of EF appeared at 7:00 LT, and the second peak was between 15:00 to 20:00 LT.This may be because more particles consume small ions, and then reduce the air conductivity and enhance EF.From 12 to 14, October, the EF was far higher than that of fair weather conditions and ranged from 154 V m -1 to 1458 V m -1 with a larger fluctuation.The PM 2.5 and atmospheric visibility showed linear correlation with EF, with coefficients of 0.5 and -0.4,respectively.In summary, our results indicate that particles under polluted conditions are greatly influential towards the atmospheric electric field.

CONCLUSIONS
One-year measurement of the atmospheric electric field (EF), air ions and pollutants was employed to explore temporal variations and interactions of these parameters.The atmospheric EF was observed to be generally higher in winter than in the other three seasons.The EF in fair weather exhibited a multimodal oscillation during the day due to the sunrise effect and air vertical convection.This is consistent with the observation results at other continental stations in the northern hemisphere.
In addition to be the meteorological parameters, aerosol particles play a significant role in atmospheric EF variation by affecting air ions and air conductivity, especially in winter.In contrast, the EF fluctuates more in polluted conditions and less in cloudless and clean conditions.In polluted periods, the atmospheric EF increased gradually with increasing amounts of PM 2.5 , and roughly reflects the trend of pollution.

Fig. 2 .
Fig. 2. Diurnal variation of mean atmospheric electric field (EF) in fair weathers on the seasonal and annual scales.

Fig. 4 .
Fig. 4. Frequency distribution of hourly (a, b, c) positive and (d, e, f) negative ion concentrations in fair weathers.

Fig. 6 .
Fig. 6.Diurnal variations of mean negative and positive (a) small, (b) intermediate and (c) large air ions.

Fig. 9 .
Fig. 9. Time series of atmospheric electric field (EF), air ion concentrations and meteorological parameters from 11 to 13 October 2015.

Table 1 .
Annual and seasonal averages of electric field (EF), air ions and meteorological parameters.

Table 2 .
Daily average of electric field (EF), meteorological parameters and their correlation coefficients (R) and ANOVA (Analysis of Variance).